crg-req-to-design

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). Outsider-authored free text from the runtime-provided requirement table file (e.g., user-supplied .xlsx/.csv) is read and converted into LLM-readable strings via mcp_server.py:_generate() → scripts/req_parser.py:parse() (e.g., note/signal fields), and then used to build crg_report.txt/output summaries; there is no explicit LLM call shown, but the workflow ingests arbitrary spreadsheet text at runtime.